Monday, 11 May 2009

British Columbia decides whether to muck up its electoral system

Andrew Coyne again is stumping for changes in British Columbia's electoral law. BC dodged the bullet in 2005 but is giving the revolver another spin on Tuesday.

The best argument for Canada's shifting to Proportional Representation would be that it might weaken some otherwise pretty persistent regional voting patterns that reward regional bloc parties. BC's adopting the system could help push it to be put in place nationally. But that doesn't outweigh the costs.

Foremost among the costs of PR is that it consistently delivers minority governments. Canada's had a few of these lately, mostly because Duverger's Law doesn't hold well when there are strong regional cleavages: it predicts two-party races at the district level, but when there's strong heterogeneity across the country, you can get Bloc Quebecois/Liberal fights in Quebec, Conservative/Liberal fights in the Prairies, Conservative/NDP fights in BC, Liberal/Conservative fights in some parts of Ontario and NDP/Liberal fights in others leading to a multi-party parliament. This is by no means set in stone: Canada usually produces single party governments. But a move to PR would entrench coalition governments. Torsten Persson and Guido Tabellini have shown that PR tends to lead to increases in government spending because of the payoffs to small parties necessary for forming coalition governments.

Coyne says the STV system will be simple for voters to understand. As he says, voters just need to be able to count to five. Yeah. Here in New Zealand, we have an even simpler form of PR: Mixed Member Proportional. Voters give one tick for a local candidate to be the local MP, the other for the Party they wish to get more seats in Parliament overall. The Party Vote determines the composition of Parliament, with first call on seats going to district-elected MPs. The system's been in place since 1996, but only about half of all voters understand which vote (Party or District) does more to determine the composition of Parliament. So here all they need to do is count to two, and they can't manage it. Counting to five would be right out.

PR is great for folks who like process for the sake of process. The composition of Parliament under PR reflects proportionate support for various parties. But it's a terrible way to form a government and hold it accountable. It yields bad policy outcomes. First Past the Post, on the other hand, is procedurally ugly. The composition of Parliament can be at large variance from aggregated national voter preferences. But at least it gives voters an easy way to throw out an underperforming government: vote for the other guy. Under PR, voters can't tell who to blame for policies they don't like and don't know for whom to vote to punish that party. My bottom line on evaluating electoral systems is to look at the results they've produced elsewhere (see Persson and Tabellini, above) and to think through how easy it is for the median voter to cast a ballot turfing out the incumbent. The latter is, to my mind, the most critical feature of a democratic system. If all else fails due to voter ignorance, we might hope that retrospective voting can form a simple rule to follow: support the incumbent if you're happy, and vote for the Opposition if not. It's tough to do even that under PR.


  1. I basically agree, but I think there might be an argument in favour of PR on robustness grounds. It seems to me that by reducing the size of the minimum willing coalition, FPP might make it more likely that crisis will lead to an overblown policy response. It's hard to imagine the wild swings we had in policy in NZ in the 70s, 80s, early 90s happening under MMP...

  2. Bigger government under PR, too, other things being equal.

  3. Depends how close you are to election, Brad: farther off, the governing party worries more that an overblown response will have negative economic effects that'll translate into incumbent-punishment down the line.

  4. My main problem with PR is that seats ≠ power, which is arguably more significant than representation. (The 'Banzhaf power index' is one of way of calculating voting power from a distribution of weighted votes). Obviously it's not the case under FPP either, however it would pay to be consistent in your criticism of it. Why is winner-take-all for votes in Parliament acceptable, but not in the votes for seats to Parliament? That's not to say the other arguments for PR might make it preferable, simply that proportionality in and of itself is a weak one.

  5. Henry, you're dead right. Small parties have disproportionate power under PR. Here in New Zealand, Winston Peters' small xenophobic New Zealand First party played kingmaker after the 2005 election -- in exchange for becoming Foreign Minister and getting the baubles of office, Peters lent his support to Labour. Helen Clark then defended him to no end through a host of campaign finance corruption scandals, refusing to fire him because his sacking would mean the downfall of the government.

  6. It's not clear that small parties have a disproportionately high amount of power - some of the time they'll have more and some of the time they'll have less. In a 30/30/30/10 split, the smallest party has zero power. In a 49/49/2 split, the smallest party has equal power to the major parties. It would be interesting to analyse the distribution of seats over a large sample (although not possible for New Zealand due to MMP's relatively recent introduction) to see if some sub-majority level of seats is disproportionately powerful based on how the votes of other parties tend to be distributed. This method does ignore political factors (for example, the NZ Green Party has probably consistently had disproportionately less power than its seat count would indicate, due to its resistance to bargaining with the National Party).

    Regardless, having a high variance in political power that is determined more or less arbitrary by the overall distribution of seats seems undesirable. You're not going to get rid of it short of unanimity rule, however it does weaken the perceived relative advantages of PR over FPP.

  7. Clarification: small parties have at least as much power under PR as compared to under FPTP, and usually more. That's what I meant by disproportionate.

    We usually expect the minimum connected winning coalition to emerge, and for that coalition to include the party preferred by the median voter. Here, the swing parties are United Future, the Maori Party, and New Zealand First. National ruled out working with NZ First in the last election and signaled that they would partner with ACT (which sits to the right of National). The coalition that emerged post 2008 was broader than minimum winning (it didn't need to include the Maori Party) as Key seems to be looking to the longer term in wanting to keep Maori as a swing party rather than a long-term Labour ally.